Abstract: There were many intersection features and surface features in aero-engine casing parts. In order to solve the problems that the existing methods couldnt recognize them effectively, a hybrid feature segmentation algorithm was proposed based on heuristics and rules. Firstly, based on the attribute adjacency graph of parts, the transition feature recognition was used to simplify the topological hierarchy of features. Secondly,based on the determination of the initial surfaces of feature segmentation, a heuristic search condition was set to perform convex edge diffusion and concave edge diffusion on the feature surfaces, and the machining features were divided into convex and concave features. Thirdly, the seed surface was determined according to the topological structures and geometric information of the feature surfaces, which were used as the driving surfaces for feature combination, and the feature parameters that could represent different features were extracted to realize the rule-based feature recognition. The effectiveness of the proposed method was verified by experiments on typical aero-engine casing parts.

Key words: attribute adjacency graph, feature segmentation, characteristic parameter, feature recognition

摘要： 航空发动机机匣零件存在大量的相交特征和曲面特征，为解决现有方法无法对其进行有效识别的问题，提出了一种融合启发式与规则的混合式特征分割算法。首先，基于零件的属性邻接图，利用过渡特征识别简化特征拓扑层次,在确定特征分割起始面的基础上，建立启发式搜索条件对特征面进行凸边扩散和凹边扩散，将加工特征分割为凸特征和凹特征两类；进而根据特征的拓扑结构及特征面的几何信息确定种子面，以其作为驱动面进行特征合并，并提取出能够表示不同特征的特征参数，据此实现基于规则的特征识别。以典型航空发动机机匣零件进行实验，验证了所提方法的有效性。

关键词: 属性邻接图, 特征分割, 特征参数, 特征识别